Keyboard shortcuts

Press ← or → to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Micro-Benchmarks & Performance

Performance, low latency, and zero overhead are fundamental design pillars of rust-recursive-deps-updater (rrdu). Because the tool interacts with large monorepos containing dozens or hundreds of crates, core algorithms must scale with strict predictability and minimal memory allocation.

This chapter details the methodology, benchmark harness, and execution workloads implemented in Criterion across parametric sample sizes of 10, 20, 50, 100, 500, and 1,000 items.

Tooling & Methodology

Micro-benchmarks are implemented using criterion (v0.8.2) under benches/rrdu_benchmarks.rs.

  • 100% Safe Execution: The benchmark harness strictly enforces #![forbid(unsafe_code)] at its root.
  • Statistical Rigor: Criterion executes warmup iterations, statistical sample collection, outlier filtering, and distribution analysis (mean, median, standard deviation).
  • Parametric Scaling: Workloads are parameterized across uniform sample sizes of $N \in {10, 20, 50, 100, 500, 1000}$ to profile asymptotic time complexity ($O(1)$, $O(N)$).
  • Zero External Network Dependencies: All benchmarks operate on in-memory ASTs, mock registries, and synthesized dependency trees, guaranteeing repeatable, isolated measurements without network latency or external side effects.

Core Benchmark Scenarios

The benchmark suite profiles four performance-critical subsystems:

1. SemVer Comparison & Classification

  • Target Functions: DependencyEntry::check_if_latest and DependencyEntry::get_dependency_row_valus in rrdu::workspace::project.
  • Workload: Evaluates Cargo SemVer compatibility rules across synthetic dependency entries with realistic version bump patterns:
    • Post-1.0 breaking updates (e.g. 1.2.0 vs. 2.0.0 -> [Need manual migration])
    • Post-1.0 non-breaking updates (e.g. 1.2.0 vs. 1.3.0 -> [No migration needed])
    • Pre-1.0 breaking updates (e.g. 0.2.1 vs. 0.3.0 -> [Need manual migration])
    • Pre-1.0 patch updates (e.g. 0.2.1 vs. 0.2.2 -> [No migration needed])
    • Up-to-date versions (e.g. 1.0.0 vs. 1.0.0)
  • Complexity: $O(N)$ linear scaling relative to total scanned dependencies.

2. Terminal Pagination & UI Line Formatting

  • Target Functions: Terminal::prepare_paginated_input_line and Terminal::get_selection_line in rrdu::interactivity::terminal.
  • Workload: Simulates interactive arrow-key navigation through large lists of $N$ items. Evaluates sliding-window page boundary calculations and row rendering for the active viewport (default: 10 rows).
  • Complexity: $O(1)$ constant-time viewport rendering, ensuring responsive navigation regardless of total repository crate count.

3. YAML Configuration Roundtrip

  • Target Functions: noyalib::to_string_with_config (Serialization) and noyalib::from_str_with_config (Deserialization) with RrduConfig.
  • Workload: Evaluates in-memory encoding and decoding of complete .rrduconfig workspace structures containing $N$ project crate definitions and exclusion rules.
  • Complexity: $O(N)$ linear scaling for both serialization and deserialization.

4. Dependency Tree Metrics & Filtering

  • Target Functions: ProjectDependency::total_values, ProjectDependency::count_valid_sections, and ProjectDependency::count_supported_total_values in rrdu::enums.
  • Workload: Evaluates multi-criteria dependency counting and section traversal across dependencies, dev-dependencies, and build-dependencies.
  • Complexity: $O(N)$ linear traversal with data locality and zero-allocation predicate matching.

Running Benchmarks Locally

Smoke Test (Single Iteration)

To verify compilation and functionality of the benchmark harness without running the full statistical sampling suite:

cargo bench --bench rrdu_benchmarks -- --test

Full Benchmark Execution

To execute full Criterion statistical measurements:

cargo bench --bench rrdu_benchmarks

Criterion generates detailed HTML reports, line plots, and distribution charts under:

target/criterion/report/index.html

Interactive Benchmark Dashboard

When viewed on GitHub Pages or built locally, the full interactive Criterion dashboard generated by the benchmark suite is embedded below:

Embedded Dashboard